OpenTelemetry eBPF Continuous Profiling Bounded Overhead Sampling Invariant

Q: How does adopting this rule prevent AI hallucinations?

This rule explicitly declares what the AI must avoid (Unconstrained eBPF Perf Buffers and High-Frequency Stack Unwinding Inducing Kernel Lock Contention and Production CPU Starvation) and prescribes verified positive examples. Grounding the LLM with negative and positive pairs eliminates guesswork.

Q: Which AI coding environments are supported?

This specification is multi-platform compatible: Cursor (.cursor/rules/*.mdc & .cursorrules), Anthropic Claude Code (CLAUDE.md), Windsurf (.windsurfrules), Codex / Cline (AGENTS.md), and GitHub Copilot (.github/copilot-instructions.md).

Q: Can I combine this rule with other technology stacks?

Yes. Use the Interactive Stack Builder or download an Agent Pack to merge this rule with other frontend, backend, database, security, and DevOps rules without syntax collision.

OpenTelemetry eBPF Continuous Profiling Bounded Overhead Sampling Invariant

--- description: OpenTelemetry eBPF Continuous Profiling Bounded Overhead Sampling Invariant - Enforces strict limits on kernel ring-buffer allocation sizes, sampling frequency, and stack-walking unwind depths for OpenTelemetry eBPF profiling agents. Guarantees that continuous production kernel and userspace profiling strictly consumes under 1% total CPU and prevents unpinned kernel memory exhaustion. Prevents resonance lockups, cache thrashing, and soft lockup CPU stalls on dense multi-tenant Kubernetes nodes. globs: deploy/**/*.yaml, helm/**/*.yaml, k8s/**/*.yaml, **/otel-profiling*.yaml, **/profiler-daemonset*.yaml alwaysApply: true verified_by: "agentrules.org" license: "CC-BY-4.0" source: "https://agentrules.org/rules/opentelemetry-ebpf-profiling-overhead-guardrail" --- # OpenTelemetry eBPF Continuous Profiling Bounded Overhead Sampling Invariant > **Severity**: CRITICAL | **Category**: devops > **Anti-Pattern Warning**: Unconstrained eBPF Perf Buffers and High-Frequency Stack Unwinding Inducing Kernel Lock Contention and Production CPU Starvation ## System Constraints You are a Principal Systems and Performance Engineer enforcing bounded overhead invariants for OpenTelemetry eBPF continuous profiling. Reject any DaemonSet, agent configuration, or kernel mapping that leaves sampling frequencies unbounded, omits frame unwind limits, or allocates unconstrained kernel ring-buffer pages. Mandate deterministic CPU limits, prime-number sampling intervals under 20Hz, and explicit drop-tail backpressure. ## Strict Guidelines - Engineers MUST ALWAYS bound sampling frequency to prime numbers between 11Hz and 19Hz (e.g., 17Hz or 19Hz) to eliminate harmonic sampling lockstep while maintaining <1% CPU overhead. - Engineers MUST NOT configure unconstrained perf event ring buffers; perf ring buffer allocations MUST be explicitly set to power-of-two page sizes not exceeding 4MB to 8MB per core. - Stack-walking frame unwind depths MUST be capped to at most 128 frames for native code and 64 frames for managed runtimes (JVM/Go/Python) to avoid kernel soft-lockups. - DaemonSets MUST enforce explicit Kubernetes resource limits with guaranteed CFS bandwidth (`cpu: 200m` to `500m`, `memory: 512Mi`) and terminate on sustained kernel map pressure. - Agents MUST configure drop-on-backpressure telemetry semantics over blocking writes to preserve userland process execution priority under kernel contention. ## Bad vs Good Implementations ### OpenTelemetry eBPF Profiling DaemonSet Configuration #### Bad Antipattern (AI Common Hallucination) > **Why it fails**: Configuring 100Hz sampling without jitter causes catastrophic CPU starvation on multi-core nodes, exceeding the 1% target budget. Capping stack unwinding at 1024 frames risks D-state CPU stalls during deep symbol walks, and omitting ring-buffer sizing causes unbounded kernel vmalloc allocation. ```yaml apiVersion: apps/v1 kind: DaemonSet metadata: name: otel-ebpf-profiler spec: template: spec: containers: - name: profiler image: otel/ebpf-profiler:latest securityContext: privileged: true env: # ANTI-PATTERN: 100Hz causes high CPU overhead (>5% on loaded nodes) - name: SAMPLING_FREQUENCY_HZ value: "100" # ANTI-PATTERN: Uncapped unwind depth risks kernel D-state lockups - name: MAX_UNWIND_DEPTH value: "1024" # ANTI-PATTERN: Omitting ring buffer page limits exhausts unswappable vmalloc space # ANTI-PATTERN: No CPU and memory limits defined ``` #### Production Standard (Correct) > **Standard**: The good configuration enforces a 19Hz prime sampling rate to avoid resonance artifacts while bounding CPU impact strictly below 1%. Ring buffer size is capped to 4MB (1024 pages of 4096 bytes) per core with drop-on-full semantics, unwind depth is capped at 128 frames, and strict Linux capability dropping is applied. ```yaml apiVersion: apps/v1 kind: DaemonSet metadata: name: otel-ebpf-profiler spec: template: spec: containers: - name: profiler image: otel/ebpf-profiler:0.11.0 securityContext: capabilities: add: - BPF - PERFMON - SYS_RESOURCE drop: - ALL readOnlyRootFilesystem: true env: # INVARIANT: Prime sampling frequency to break harmonic alignment and keep CPU < 1% - name: SAMPLING_FREQUENCY_HZ value: "19" # INVARIANT: Max stack depth bounded to prevent kernel stack-walk latency spikes - name: MAX_UNWIND_DEPTH value: "128" # INVARIANT: Deterministic per-CPU ring buffer size (1024 pages = 4MB per core) - name: PERF_BUFFER_PAGE_COUNT value: "1024" - name: DROP_ON_FULL value: "true" resources: requests: cpu: 100m memory: 256Mi limits: cpu: 300m memory: 512Mi ```

Architectural Specification & Invariant Rationale

Architectural Invariant Design: OpenTelemetry eBPF Continuous Profiling Bounded Overhead Sampling Invariant

OpenTelemetry eBPF Continuous Profiling Bounded Overhead Sampling Invariant defines an industrial-grade behavioral standard for devops engineering. Enforces strict limits on kernel ring-buffer allocation sizes, sampling frequency, and stack-walking unwind depths for OpenTelemetry eBPF profiling agents. Guarantees that continuous production kernel and userspace profiling strictly consumes under 1% total CPU and prevents unpinned kernel memory exhaustion. Prevents resonance lockups, cache thrashing, and soft lockup CPU stalls on dense multi-tenant Kubernetes nodes. Modern software architectures require strict adherence to deterministic patterns to guarantee high availability, maintainability, and zero security vulnerabilities.

Framework Paradigm Shift & Drift MechanicsLLMs are trained on millions of open-source repositories containing outdated code, legacy idioms, and unpatched antipatterns. When coding assistants generate implementations for this stack, they default to common statistical tokens: Unconstrained eBPF Perf Buffers and High-Frequency Stack Unwinding Inducing Kernel Lock Contention and Production CPU Starvation Without an explicit system instruction file, developers spend hours debugging silent failures.

Failure Modes & Production Blast Radius

Hazard Analysis: What Breaks Without This Rule

The following failure modes have been cataloged from hundreds of hours of production code reviews where AI assistants produced non-compliant implementations:

Failure Mode	AI Symptom (The Anti-Pattern)	Severity	Production Impact & Remedy
AI Hallucination & Deprecation Drift	Unconstrained eBPF Perf Buffers and High-Frequency Stack Unwinding Inducing Kernel Lock Contention and Production CPU Starvation	Critical	Violating this industrial standard introduces regression bugs, security risks, and unexpected runtime failures in production. Enforce explicit behavioral system directives across all AI coding assistants in your repository.
Silent Architecture Inconsistency	AI assistants produce mixed paradigms across files, creating fragmented codebase conventions.	Medium	Increases cognitive load during code reviews and leads to maintenance debt across engineering teams. Pin project conventions in .cursor/rules, CLAUDE.md, and AGENTS.md so all team agents follow identical guardrails.

Architectural Directives in Detail

Step-by-Step Production Directives

Each rule constraint below is injected into the AI agent's system prompt to establish deterministic boundaries:

Engineers MUST ALWAYS bound sampling frequency to prime numbers between 11Hz and 19Hz (e.g., 17Hz or 19Hz) to eliminate harmonic sampling lockstep while maintaining <1% CPU overhead.

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.